The American College of Obstetricians and Gynecologists (ACOG) Research Department mailed questionnaires to 1,000 randomly selected members of the Collaborative Ambulatory Research Network (CARN), a group of practicing obstetrician-gynecologists representative of ACOG Fellows relative to geographic location, age, and sex and who have agreed to participate in periodic ACOG surveys.^22,23 The participant sample excluded physicians listed in the CARN database as practicing gynecology only. There were a total of 5 mailings, with each subsequent mailing after the first one being sent only to physicians who had not yet responded. Participants were not offered any compensation for returning a survey.

From the original 1,000 sampled physicians, 584 returned surveys; however, 74 of the respondents no longer treated pregnant patients and were excluded from the sample, resulting in a response rate of 55.1% (510 completed questionnaires from a total possible of 926). There was no difference in response rate between men and women (54.9% vs. 56.3%, p = 0.653); however, nonrespondents were younger than respondents (47.2 ± 0.4 years vs. 49.0 ± 0.4 years, p = 0.003).

The survey contained questions regarding physician and patient population demographics, physician practice patterns related to treatment of pregnant diabetic women, and opinions concerning barriers to appropriate treatment. The questionnaire was evaluated using a Question Appraisal System (QAS) developed by RTI International to evaluate the structure and effectiveness of the questionnaire form itself. The QAS was based on several previous question appraisal systems and, in part, from a method developed to examine and classify the cognitive processes inherent in the question-answering process.²⁴ Responses were entered into a software package (SPSS 16.0, SPSS, Inc., Chicago, Illinois) data file for analysis. The study was approved by the institutional review board of RTI International.

We used cluster analysis on physician reports of their patients’ characteristics to develop a dichotomy of practices of low minority patient populations with a high proportion of insurance coverage and those with high minority patient populations and low insurance coverage. Maternal-fetal medicine (MFM) practices (48 practices) were not included in this dichotomy, as we a priori expected their characteristics to differ from non-MFM practices. Among the 462 non-MFM respondents, 304 physicians were classed in the low-minority/high-insurance group, 138 in the high-minority/low-insurance group, and 20 could not be classified because they did not answer one or more of the cluster-defining questions. Results are presented separately for physicians in low-minority/high-insurance practices, physicians in high-minority/low-insurance practices, and MFMs. Statistical comparisons were made across all 3 groups and, where appropriate, for the pair-wise comparisons.

We report values for continuous variables as mean ± standard error of the mean (SEM) and frequencies in percent. We use F-tests for differences between means, Mann Whitney U test for scaled responses to test for differences between 2 groups, and the Kruskal Wallis test for differences among 3 groups. For categorical data we used χ² tests. Statistical significance was set at p < 0.05.

Among the 510 eligible respondents, 452 (88.6%) practiced obstetrics and gynecology, 48 (9.4%) were MFM specialists, 4 (0.8%) practiced obstetrics only, and 6 (1.2%) listed other as their specialty. Practice setting differed among MFMs and the 2 patient demographic practice groups (p < 0.001), with MFMs more likely to practice in an academic setting, physicians with low minority/high insurance patient populations more likely to be in private group practice, and physicians with high minority/low insurance patient populations more evenly spread among the practice types. Physicians with high minority/low insurance patient populations were more likely than the other 2 groups to have a solo practice (Table I). The mean number of years in practice (17.5 ± 1.5) did not differ among the 3 groups (p = 0.587). Physicians with high-minority/low-insurance patient populations reported on average performing more deliveries per year (160 ± 10, p = 0.023) as compared to physicians with low minority/high insurance patient populations (134 ± 5) or MFMs (129 ± 20). Patient demographics and insurance status were similar between MFMs and physicians with high minority/low insurance patient populations (Table I).

In general, the reported care provided to diabetic women by the responding physicians is in line with ACOG recommendations. Most physicians asked their patients about their pregnancy plans always or most of the time. Many of the responding physicians report that they always counsel their diabetic patients considering pregnancy regarding glycemic control, vitamin use, their prescription medications, and weight, diet and exercise. If the patient is pregnant, the proportion of physicians who always counsel is > 90% for all but exercise (82.7%). It is perhaps somewhat disappointing that about 1 in 5 responding physicians do not routinely counsel pregnant diabetic women about exercise. In addition, 14–21% of the responding physicians do not routinely counsel their diabetic patients about glycemic control, vitamin use, their prescription medications, and weight, diet and exercise if the patient does not express a desire to get pregnant. Considering that almost half of pregnancies in the U.S. are unplanned²⁵ and that the responding physicians themselves believe that, on average, < 60% of the pregnancies among their patients are planned or intended, it is concerning that physicians are not routinely counseling their patients with diabetes on these issues. Planning their pregnancies and using effective contraception until glycemic control is achieved are important strategies for physicians to encourage in women with diabetes in order to reduce the risk for birth defects in their infants and other adverse pregnancy outcomes.

It was also disappointing to find that, although virtually all physicians rated conversion to type 2 status as a concern at some level for their GDM patients, less than half of these physicians fully evaluated their GDM patients’ glucose metabolism postpartum by performing an OGTT. Prevention of or delay in conversion to type 2 diabetes can reduce the risk of potential complications of diabetes in the woman and of birth defects in any future infant.¹² Physician concern regarding the future consequences of a GDM pregnancy does not appear to be matched by appropriate screening, counseling and referral practices. This would appear to be an area where practice can improve.

There are several limitations to this study. The physicians in this sample are all members of CARN, a voluntary research network. ACOG sends out periodic invitations to subgroups of Fellows to join CARN in order to keep CARN membership similar in age, sex, and geographic representation to all actively practicing ACOG Fellows. Several surveys a year are sent to both CARN members and randomly selected Fellows. Over the 18-year history of the CARN,^22,23 randomly selected Fellows rarely differed from CARN members in their responses; however, we cannot exclude the possibility that CARN members may differ from other ACOG Fellows in ways we have as yet been unable to measure. Another limitation of the study is that the data are all derived from physician self-report. Finally, although the response rate was comparable to past CARN survey studies, > 40% of eligible participants declined to respond. Nonrespondents were younger on average, implying less clinical experience. It is possible that the nonrespondents were less knowledgeable regarding appropriate practice for pregnant diabetic women, in which case our results are overly optimistic.

It is not surprising that MFM specialists might practice differently from general obstetrician-gynecologists. MFMs likely see a higher percentage of pregnancies complicated by diabetes of all types. The differences between MFMs and the other ACOG Fellows in this study are consistent with MFMs being more aware of adverse outcomes associated with diabetes in pregnancy and of best practices for management. However, although a higher proportion of MFMs responded that they ask about their diabetic patients’ pregnancy plans and provide preconception counseling, many MFMs rarely see nonpregnant patients but rather are more likely to treat high-risk pregnant women referred to them. Thus this more conscientious practice by MFMs may not translate into increased quality of preconception care received by diabetic women.

The differences between physicians based on their patient demographics are of interest. Physicians with high minority/low insurance patient populations appeared to practice more conscientiously in certain aspects of care provided to diabetic women. In general, these physicians were more likely to manage their diabetic patients’ blood glucose personally and to counsel their patients regarding appropriate practices. They also appeared to have a higher level of concern regarding their patients’ access to resources (e.g., referral to specialists, access to healthful food, and access to educational resources) and regarding their patients’ ability and willingness to comply with provider recommendations. Their practice and opinion profile was generally closer to that of MFMs than to the physicians with low minority/high insurance patient populations (Tables I, II, and IV).

Unfortunately, we do not have data to assess possible explanatory causes for these differences in relation to patient population. Our data are consistent with physicians with high minority/low insurance patient populations having access to fewer resources (e.g., less likely to practice in a group practice, on average perform more deliveries, and are more likely to manage glucose control for their patients without the assistance of a diabetes specialist) and to perceiving that their patients have less access to appropriate resources. The more precarious health insurance situation for these patient populations on average certainly seems relevant to understanding the different level of concern by these physicians. The self-reported behavior of these physicians indicates that pregnant diabetic women without insurance but with access to an obstetrician are receiving quality care; however, the physicians and their patients may lack infrastructure and access to resources.

A majority of all the responding physicians rated limitations on the amount of time spent with patients and, perhaps related, problems with reimbursement and insurance coverage as barriers to quality care. Among physicians with a high proportion of vulnerable patients (minorities and underinsured) a majority rated patient compliance as a barrier (Table IV). Educational resources for patients can help the physician increase awareness and knowledge of diabetes among women with or at risk for diabetes and of the strategies and practices to manage and ameliorate the harmful consequences. In this study many physicians, especially those who serve vulnerable populations, expressed concern over a lack of such educational resources.

In summary, diabetic women with access to an obstetrician generally receive quality care regardless of health insurance status. Postpartum evaluation of glucose metabolism in GDM patients appears to be an area that could be improved. A majority of responding physicians considered the amount of time spent with patients and problems with reimbursement or insurance coverage to be barriers to quality care. Women in patient populations with high minority and/or low insurance status may lack access to resources and educational materials.